Meta reportedly tracking employee keystrokes to train AI systems

Meta is reportedly turning to an unconventional source of training data for its artificial intelligence systems: its own employees. According to a report first published by Reuters, the company plans to collect data from staff's mouse movements, keystrokes, and on-screen interactions as part of efforts to develop more capable and efficient AI models.

The development highlights the growing competition among tech companies to secure new types of training data, which remains a critical resource for improving modern AI systems. As traditional datasets become more limited or regulated, firms are increasingly exploring alternative sources of behavioural and operational data to enhance model performance.

A Meta spokesperson issued a statement explaining the initiative, saying that if the company is building AI agents designed to help users complete everyday computer-based tasks, the systems need real-world examples of how people actually interact with software interfaces. This includes actions such as moving a mouse, clicking buttons, and navigating dropdown menus.

The spokesperson added: "If we're building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them — things like mouse movements, clicking buttons, and navigating dropdown menus. To help, we're launching an internal tool that will capture these kinds of inputs on certain applications to help us train our models. There are safeguards in place to protect sensitive content, and the data is not used for any other purpose."

The reported initiative reflects a broader trend in the AI industry, where companies are increasingly relying on unconventional, previously unused data sources to train models. In recent weeks, reports have also highlighted how older startups and enterprise software platforms are being repurposed to extract internal communications, such as Slack archives, Jira tickets, and other workplace data, for AI training.

Industry observers say these developments raise new privacy considerations, particularly as organisations expand data collection methods beyond traditional datasets to include real-time human behavioural tracking in workplace environments.